Podcast: Artificial intelligence in payment integrity
As health plans look to artificial intelligence (AI) and machine learning to improve their operations and payment integrity programs, they must understand how it can be practically applied. But payers also want to be reassured the technology will be adopted appropriately and responsibly—and won’t increase provider abrasion.
On the fifth episode of Cotiviti’s Payment Integrity Insights podcast, Cotiviti’s Brett Arnold, senior vice president of product development, is joined by Anandhi Periyanan, senior vice president of R&D, to discuss the role of AI in payment integrity. Listen and learn:
- How Cotiviti uses AI responsibly to increase payment integrity value
- How healthcare payers are using AI and the challenges they’re facing
- How to separate fact from fluff when evaluating AI partners
Podcast guests
![]()
|
Brett Arnold Senior Vice President, Product Development |
![]()
|
Anandhi Periyanan Senior Vice President, Research and Development |
Don’t miss this opportunity to learn how AI can be appropriately implemented to improve payment integrity to benefit the entire healthcare ecosystem. Stay tuned for part two of the podcast next week as we focus on four key tenets for incorporating AI into your payment integrity program.
Podcast transcript
Brett: In this series, we'll discuss how the market is being flooded with promises of what AI can do for healthcare, how you can separate fact from fluff, and how safe and purposeful application of AI can bring real value to health plans trying to thrive in today's market. In part one today we'll focus on the buzz in the marketplace, a retrospective on AI's potential to help health plans meet their goals, and our vision of how AI can be used safely and effectively to improve the value health plans realize from their payment integrity efforts.
AI is a key investment area for payment integrity. Generative AI capabilities offer promise to significantly improve effectiveness of health plan operations and payment integrity programs. But understanding what AI applications are truly useful in adding value to these programs and what responsible AI looks like is key to plans getting the most value possible. I'm joined today by Anandhi Periyanan, Cotiviti senior vice president of research and development. Anandhi and I work together to improve our products and services for health plans leveraging technology. And more recently, we've been spending a lot of time focusing on artificial intelligence and how we can use it to responsibly improve our services.
Before we get into AI and payment integrity, let's take a moment to talk about Cotiviti. Who is Cotiviti and why are we qualified to talk about AI and payment integrity? We've been partnering with health plans to ensure payment accuracy for over 20 years. We work with more than 100 health plans in payment integrity for these clients. We identify previously unrealized medical administrative cost savings with solutions, reviewing all major claim types, repayment and post-payment to help ensure accuracy, determine responsibility, and detect patterns of fraud, waste, and abuse.
Recently, Cotiviti took time to clearly define our vision for how AI should be used to further the mission of payment integrity . Anandhi, you had a lot to do with developing that vision. Can you take just a moment to tell us about that?
Anandhi: Cotiviti's vision is to enable a high quality and a viable healthcare system. We embrace the use of AI in support of our mission. This means we do not have a different vision for Cotiviti using AI. We are using AI to drive our existing mission and mission. We leverage our expansive informatics infrastructure and subject matter expertise. We use AI and related technologies to improve and expand the solutions we provide to our clients, including the increase of the value delivery, decreasing the client administrative burden, improving the results, accuracy, and enhancing our client service experience. At Cotiviti, this means we are not approaching AI as a new commercial product. We use and view it as an exciting new tool to improve the products and services we already have and develop new products and bring it to the market. We deploy advanced AI in an augmented intelligence approach to support the expert decisions with relevant information and proprietary insights.
Perhaps most importantly, we will strategically limit AI automation to enabling our human specialists to improve performance and client experience. At Cotiviti, this means that we are not replacing human clinical decision-making or judgment. We will use AI to prepare the clinical content, but not to make the findings decision. We hear occasionally where healthcare organizations have made AI a decision-maker and we are intentionally avoiding that situation.
Brett: So let's dive into our content today. As promised, we'll talk about the potential of AI in payment integrity. We'll do this first by providing some context on the healthcare market and on AI. We'll delve into how we believe AI can add value to payment integrity, sharing examples, and bringing home our overriding beliefs.
First off, AI is a tool that will add value. It's not a solution on its own. Second, AI should be used to improve your results, the results that matter for your customers. Third, AI must be used responsibly and fourth, AI does not replace human expertise. It augments it. So everyone is aware that the buzz around AI increased significantly in 2023. This is largely from the release of ChatGPT from OpenAI, which is both extremely exciting and a little scary. While AI and underlying technologies have been in development and in use for quite some time, we are now seeing the capabilities become core to most companies.
There seems to be strong consensus building that AI, like any good new technology, will evolve your workforce. It won't replace humans, but it will likely redistribute how you employ them. AI will also increase productivity. Your team, everyone should be thinking: What would you do with 20% more time? That's a question you should be asking yourself in considering how to incorporate generative AI.
Finally, a McKinsey study that I recently read showed that market leaders, they're really seeing growth from AI. Those that are really getting a bang from AI are using it to focus on top line growth, improving their products and services for their customers. So while it's tempting to get caught in the trap of AI as an efficiency tool or an efficiency play, those that are really using AI to lead in the market are focusing AI on top line on improving their products and services for their customers.
At Cotiviti, we are clearly hearing from health plans that they're excited about AI and how it can transform their programs. Our clients are at different points in their maturity and are trying to balance where they can develop their own AI capabilities and where they would benefit more from partnering with folks like Cotiviti, all of them are working to understand how governance works for AI and looking for leaders to help them determine that path.
When I talk to most people, including myself, they're both excited and a little bit scared of AI. Personally, I find AI very exciting in healthcare. I recently read a book called The Age of Scientific Wellness by Leroy Hood and Nathan Price, and it explains how we can use AI to really revolutionize medicine. The book says AI is a key component to detect and predict the transition from wellness to disease. Just like doctors have always researched for patterns in smaller data sets where they work, AI will seek patterns in infinitely more complicated data sets that exist now and in the future. So the result and what makes me excited is that we're changing healthcare instead of from treating disease to detecting and preventing the transition to disease. The book calls this delivering healthcare that is predictive, preventative, personalized, and participatory. Very exciting.
I also heard a presentation from McKinsey recently, which showed the substantial potential to create value for the healthcare system from AI and for all of us as consumers. They quoted $200–$360 billion potential net savings in healthcare spend from AI and an additional $150–$260 billion from generative AI. So in total, $350–$600 billion in potential savings and cost savings. I don't know about you, but healthcare is expensive and we can use that help. So there's a lot to be excited about in AI.
But on the flip side, AI is also scary. I've heard many people talk about AI takeover, when “Skynet” becomes self-aware. So that's one fear. Maybe a more practical short-term fear is increasing bias. I think we all know that the systems and the processes and the data that we're producing as a society have bias in them. So will AI accelerate that bias and make it worse, or can it be used to help eliminate it? That's a real question. And also mass unemployment. Can AI make us so efficient that a lot of people will lose their jobs?
I personally don't think that will be the case. I definitely think there will be a shift in employment how you utilize resources. But another favorite quote of mine that I've heard around the AI circuit is that AI won't take your job, but someone who knows how use AI will. So I think this is a train that's kind of already left the station and it will make us more productive. The question is, what are you going to do with that productivity? How are you going to help increase value for your clients or for your business? That's where we should be focusing, and when you think about AI and how it impacts your job.
So let's provide a little more foundation here. We're talking about AI a bunch. What even is AI? Where is it headed? Anandhi, you’re an R&D and engineering leader. Can you break it down for the layperson?
Anandhi: A quick background on AI: definition and the context of its technology. The artificial intelligence community divides its capability and functionality. So based on today's conversation, we just explained the three types of capabilities of AI, the narrow AI, general AI, and super AI. Narrow AI is designed and trained for a specific task, narrow range tasks or purpose. These narrow systems perform their designated tasks, but mainly lack the ability to generalize the tasks. Personal virtual assistants like Alexa or Siri, image recognition software and other language translation tools are examples of narrow AI. General AI, it refers to the AI system that have human intelligence and can perform various tasks. Systems have the capability to understand, learn, and apply across a wide range of tasks that are similar to a human brain. In general, the general AI remains a theoretical concept and no AI has achieved this level of intelligence yet, but it is in the works.
The last but not the least, super AI is also known as super intelligent AI that surpasses the intelligence of human in problem solving, creativity and overall ability. Super AI has the ability to develop emotions, desires, needs, and beliefs of their own, and they can make their own decisions and solve their own problems. The recent buzz is generative AI. The generative AI is the type of AI technology that can produce various types of content, including text, images, audio, and synthetic data. The trend has been driven by the simplicity of a new user interface for creating high-quality text, graphics, and videos in a matter of seconds.
Brett: So let's talk about how AI progression is accelerating. So with any new technology, there is a hype cycle, right? Where there's more buzz and more news about a technology than there is actual value can create. For the narrow AI that Anandhi described, we're definitely beyond that hype cycle. Most companies now have gotten through that and determined where are places where machine learning or natural language processing more narrow AI components can help their business. A lot of the buzz presently is generated by generative AI, as we mentioned, with large language models led by ChatGPT getting into the real world, getting a huge number of users in the last 18 months. But it's not just that. Also the scaling, right—artificial intelligence as a concept's been around a long time, but we're just getting to where the data and the technology and the infrastructure can really use it at scale.
So that ability for the technology to catch up to the theory has just happened in spades here recently as well. So on Cotiviti’s front, we're certainly past the hype cycle on narrow AI and use machine learning, natural language processing, and other artificial intelligence, the narrow AI for many business purposes, but we're still relatively early working on pilots and incorporating generative AI cautiously into our operation, and we're not alone on this front. According to a recent survey from McKinsey, over 70% of healthcare organizations are either using or planning to use cutting edge AI technologies like generative AI presently. So it seems like most companies are in that exploratory or early usage phases for generative AI.
Anandhi: And in the same study, McKinsey considered the potential for AI to assist in payment integrity. The opportunity is only on the increase. AI and generative AI can be used to enhance the efficiency of claim review, editing, and recovery as shared in the Cotiviti client conference. The impact potential, I'm just going to lay out some statistics. 8–30% of PI admin cost can be automated with AI, and 0.4–1.7% of medical cost that goes to claim review, editing, and recovery can be decreased. It will gain more program value by requesting fewer medical records, more accuracy, and avoiding adjustments that will help Cotiviti to go deeper into the right claims.
Generative AI offers enormous productivity benefits for individuals and organizations and businesses are forging ahead and exploring how technology can improve their internal workflow and enrich their products and services. According to the research by McKinsey, one third of the organizations are already using generative AI, and at least one business function industry analyst, Gartner, projects that more than 80% of the organizations will have deployed generative AI or used generative AI application programming interfaces or APIs by 2026, which is two short years from now.
Brett: Yeah, that's not long. So the opportunity is great for generative AI, which is exciting. However, the buzz is also great, and sometimes it's tough to know what's real and what's just marketing. So there's a lot of companies out there talking about AI and you should definitely listen to them, but be careful to ask yourself, and maybe more importantly, ask them a few questions. So many companies are hyping AI are either startups and they're looking to get a foothold and looking to learn from you. If they don't have expertise in your business, then you are their expertise. The real magic and value from AI comes when you're bringing together data scientists and experts. Consider that if they're someone who has expertise and has the experts to bring, then they're bringing expertise to you. Otherwise you're bringing it for them and training them. Second, make sure you are looking at the inputs that a company has for AI.
AI needs data to learn from, right? AI does not make up new things. It's looking at past results, past data and needs lots of it to train in order to create future value. So if a company doesn't have a significant history in solving a problem, then they're looking at your history, right? They're looking for your data to train them on how that works. So ideally, they would have a broader data set even than just your data as an input to their AI. Then they're bringing value that you can't get from your own data. It's just your data and your input than you're not gaining learnings from the broader market. So consider their data inputs. Finally, I've also seen that companies are selling platforms, are often talking about AI as a feature in their software. On the other hand, solution providers or services companies are less likely to tout AI as a feature.
Cotiviti actually fits in this second category. We tend to communicate our results to clients instead of technology features. So we measure ourselves often by how much medical cost savings we create or how much administrative costs avoidance we can create for our clients, right? Those are the measures of success that we had before, and we all in the future, we're often saying, yes, we use artificial intelligence, but we're using it to create an increase in this metric or this measure of success. So just consider, is it a software vendor selling a feature that will help you, or is it a services vendor that you should be asking more about? What's the value you're getting from artificial intelligence? So the key point here is make sure you're asking about the specific results that matter to you. Don't ask, are you using AI? Ask, how much will AI improve my results and how are you going to measure and report that improvement to me?
At the end of the day, I'd be wary about a company that talks too much about AI or is calling themselves an AI company. AI is a great technology, it's very exciting, but it is just a technology. If they're only an AI company, then what are they going to be in five years? In five years, AI won't be a buzzword anymore. It's like the internet was 20 years ago, or cloud computing. Can you imagine companies saying they're an internet company today? I mean, of course you're an internet company. Everybody's an internet company. That's how it's going to be with AI. So be asking, what are they really? Because if they're just an AI company, where are they going? So let's get more specific to payment integrity at health plans. I mentioned that AI is now core to market-leading companies. What about health plans? Where are they in the journey of incorporating AI and generative AI?
Anandhi: You and I, we had many conversations with our clients about AI, and answers certainly vary across our clients. In some cases, they're well into the journey on AI and in others, they're just beginning On the whole, I see a few themes. Some of our clients are moving forward slowly. As with any new technology, they're struggling for resources and making their case with their management for investment. Often this means they're looking to their vendors and partners to lead them in incorporating AI. Vendors like Cotiviti can invest in exploration across clients and market and help drive this type of change. Clients are looking at many of the same initial payment integrity use cases at Cotiviti. For narrow AI, this means improving algorithms to detect fraud, waste, and abuse, or to improve selection of cases for review. For generative AI, this could include the smart search, text summarization and medical record summarization, interpretation, and auditing.
In terms of governance, clients are at varying points of AI maturity. Some have an AI governance committee and some are bringing it, and all of them want vendors to be good partners and help them figure out what responsible AI should look like. In summary, our clients are mixed and implementing slowly. They want to see the AI used for value. They need real applications, not fluff or spin, embedded in the workflow to both increase the value and also to increase productivity. That's where we get the inspiration for the value we embed in our solutions at Cotiviti.